Date: Mon, 25 Nov 2019 22:35:18 -0800 From: Chris <bsd-lists@BSDforge.com> To: Artem Viklenko <artem@viklenko.net> Cc: <freebsd-net@freebsd.org> Subject: Re: How to remove watchdog? Message-ID: <a0793bc91ef3ab05a8ac554d023c0451@udns.ultimatedns.net> In-Reply-To: <a326ce0a-2d9c-a357-13c2-9d027b933dc8@viklenko.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 26 Nov 2019 07:55:11 +0200 Artem Viklenko artem@viklenko=2Enet said > Hi! >=20 > I have several small boxes with realtek nics acting as a router/firewall=2E > Also had same issues=2E FreeBSD driver didn't work at least for me so I=20 > switched to Realtek's driver=2E But after some time traffic stops passing= =20 > my routers=2E Did some investigation and found that the issue is 9k mbufs=2E > As far as I understand more traffic you push more issues with 9k mbufs > appears due to memory fragmentation=2E > You can check it with 'vmstat -z | grep mbuf'=2E Hah! I wondered about that=2E It *appeared* to be due to increaded load on the NIC -- hence our interest in upping to 10G NICs=2E >=20 > So I decided to do wery dirty hack - I've changed Jumbo_Frame_9k > to Jumbo_Frame_4k in the if_re=2Ec from Realtek's latest 1=2E95 driver=2E > It comiles and work on FreeBSD 10=2Ex and 11=2Ex and now it just works > (vendor says that driver is for older versions of the FreeBSD OS)=2E > And no more issues=2E >=20 > ITEM SIZE LIMIT USED FREE REQ FAIL SLEEP >=20 > mbuf_packet: 256, 2362080, 2, 1263, 2054916, 0, 0 > mbuf: 256, 2362080, 514, 1776,3460790080, 0, = 0 > mbuf_cluster: 2048, 369076, 1265, 31, 154081, 0, 0 > mbuf_jumbo_page: 4096, 184537, 513, 294,1592339809, 0, = 0 > mbuf_jumbo_9k: 9216, 54677, 0, 0, 0, 0, 0 > mbuf_jumbo_16k: 16384, 30756, 0, 0, 0, 0, 0 >=20 > Now driver use mbuf_jumbo_page not mbuf_jumbo_9k and no fails=2E >=20 > I'm ok with mtu 1500 in my environment and I don't know if mtu 9000 will > work with this change=2E But at least it is stable now even after 100 days > of uptime (just rebooted after upgraded to 11=2E3-RELEASE-p5)=2E Thank you *very* much, Artem=2E I'll have a closer look=2E I'm thinking of taking your concept, and upping it to 7k=2E I'll post back, if anything good comes of it=2E :) >=20 > Hope this helps=2E It does=2E :) --Chris >=20 >=20 > 26=2E11=2E19 02:44, Chris =D0=BF=D0=B8=D1=88=D0=B5: > > Or at least make it non fatal=2E > > OK here's the story; I'm experimenting with a multiport NIC (re(4)) > > as we hope to start using multiport 10G NICs=2E > > Any of the re's we've used in the past have been very stable, which > > is why I picked the one I did for this experiment=2E This one has been > > performing rock solid for some 4 to 6 mos, under full time use=2E That > > is until the last week=2E Where we're seeing: > > watchdog timeout > > repeated frequently=2E Which is ultimately fatal=2E ifconfig up/down will > > not resuscitate it=2E Nor will service ifconfig restart, or plugging/ > > unplugging the cable(s)=2E Bouncing the server is the only cure=2E Which > > is unacceptable=2E Any, and All suggestions, or insight into the matter > > GREATLY appreciated=2E Note; while this is an old 11=2E1, we're not plannin= g > > to up this box until we can confirm this can be cured=2E :) > >=20 > > Details follow: > > 11=2E1-STABLE r327867 amd64 > >=20 > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > > watchdog timeout > >=20 > > rc=2Econf(5) > > ifconfig_re0=3D"inet AA=2EBBB=2ECC=2EXX netmask 255=2E255=2E255=2E0 rxcsum txcsum t= so4" > > ifconfig_re1=3D"inet AA=2EBBB=2ECC=2EWW netmask 255=2E255=2E255=2E0 rxcsum txcsum t= so4" > > ifconfig_re1_alias0=3D"inet AA=2EBBB=2ECC=2EZZ netmask 255=2E255=2E255=2E0" > >=20 > > ifconfig(8) > > re0: flags=3D8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu = 1500 > >=20 > > =C2=A0=C2=A0=C2=A0=C2=A0options=3D8219b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HW= TAGGING,VLAN_HWCSUM,TSO4,WOL_MAGIC,LINKSTATE> > > =C2=A0=C2=A0=C2=A0=C2=A0ether 00:13:3b:0f:13:44 > > =C2=A0=C2=A0=C2=A0=C2=A0hwaddr 00:13:3b:0f:13:44 > > =C2=A0=C2=A0=C2=A0=C2=A0inet6 fe80::213:3bff:fe0f:1344%re0 prefixlen 6= 4 scopeid 0x1=20 > > =C2=A0=C2=A0=C2=A0=C2=A0inet AA=2EBBB=2ECC=2EXX netmask 0xffffff00 broadcast= 24=2E113=2E41=2E255=20 > > =C2=A0=C2=A0=C2=A0=C2=A0nd6 options=3D23<PERFORMNUD,ACCEPT_RTADV,AUTO_= LINKLOCAL> > > =C2=A0=C2=A0=C2=A0=C2=A0media: Ethernet autoselect (1000baseT <full-du= plex>) > > =C2=A0=C2=A0=C2=A0=C2=A0status: active > > re1: flags=3D8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu = 1500 > >=20 > > =C2=A0=C2=A0=C2=A0=C2=A0options=3D8219b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HW= TAGGING,VLAN_HWCSUM,TSO4,WOL_MAGIC,LINKSTATE> > > =C2=A0=C2=A0=C2=A0=C2=A0ether 00:13:3b:0f:13:45 > > =C2=A0=C2=A0=C2=A0=C2=A0hwaddr 00:13:3b:0f:13:45 > > =C2=A0=C2=A0=C2=A0=C2=A0inet AA=2EBBB=2ECC=2EWW netmask 0xffffff00 broadcast= 24=2E113=2E41=2E255=20 > > =C2=A0=C2=A0=C2=A0=C2=A0inet AA=2EBBB=2ECC=2EZZ netmask 0xffffff00 broadcast= 24=2E113=2E41=2E255=20 > > =C2=A0=C2=A0=C2=A0=C2=A0inet6 fe80::213:3bff:fe0f:1345%re1 prefixlen 6= 4 scopeid 0x2 > > =C2=A0=C2=A0=C2=A0=C2=A0nd6=20 > > options=3D23<PERFORMNUD,ACCEPT_RTADV,AUTO_LINKLOCAL> > > =C2=A0=C2=A0=C2=A0=C2=A0media: Ethernet autoselect (1000baseT <full-du= plex>) > > =C2=A0=C2=A0=C2=A0=C2=A0status: active > > lo0: flags=3D8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384 > > =C2=A0=C2=A0=C2=A0=C2=A0options=3D600003<RXCSUM,TXCSUM,RXCSUM_IPV6,TXC= SUM_IPV6> > > =C2=A0=C2=A0=C2=A0=C2=A0inet6 ::1 prefixlen 128 =C2=A0=C2=A0=C2=A0= =C2=A0inet6 fe80::1%lo0 prefixlen 64 > > scopeid=20 > > 0x3 =C2=A0=C2=A0=C2=A0=C2=A0inet 127=2E0=2E0=2E1 netmask 0xff000000 =C2=A0= =C2=A0=C2=A0=C2=A0nd6=20 > > options=3D21<PERFORMNUD,AUTO_LINKLOCAL> > > =C2=A0=C2=A0=C2=A0=C2=A0groups: lo > > pciconf(8) > > re0@pci0:5:0:0:=C2=A0=C2=A0=C2=A0 class=3D0x020000 card=3D0x012310ec ch= ip=3D0x816810ec=20 > > rev=3D0x07 hdr=3D0x00 > > =C2=A0=C2=A0 vendor=C2=A0=C2=A0=C2=A0=C2=A0 =3D 'Realtek Semiconductor= Co=2E, Ltd=2E' > > =C2=A0=C2=A0 device=C2=A0=C2=A0=C2=A0=C2=A0 =3D 'RTL8111/8168/8411 PCI= Express Gigabit Ethernet=20 > > Controller' > > =C2=A0=C2=A0 class=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 =3D network > > =C2=A0=C2=A0 subclass=C2=A0=C2=A0 =3D ethernet > > re1@pci0:6:0:0:=C2=A0=C2=A0=C2=A0 class=3D0x020000 card=3D0x012310ec ch= ip=3D0x816810ec=20 > > rev=3D0x07 hdr=3D0x00 > > =C2=A0=C2=A0 vendor=C2=A0=C2=A0=C2=A0=C2=A0 =3D 'Realtek Semiconductor= Co=2E, Ltd=2E' > > =C2=A0=C2=A0 device=C2=A0=C2=A0=C2=A0=C2=A0 =3D 'RTL8111/8168/8411 PCI= Express Gigabit Ethernet=20 > > Controller' > > =C2=A0=C2=A0 class=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 =3D network > > =C2=A0=C2=A0 subclass=C2=A0=C2=A0 =3D ethernet > >=20 > > Thanks again! > >=20 > > --Chris > >=20 > >=20 > > _______________________________________________ > > freebsd-net@freebsd=2Eorg mailing list > > https://lists=2Efreebsd=2Eorg/mailman/listinfo/freebsd-net > > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd=2Eorg" >=20 > --=20 > Regards!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?a0793bc91ef3ab05a8ac554d023c0451>